ABSTRACT
In a study of 371 males with chest pain, data were collected on the concentration of plasma cholesterol and plasma triglycerides. For 51 patients there was no evidence of heart disease; for the remaining 320 there was evidence of narrowing of the arteries. Interest lies in assessing whether the two measurements made can be used to discriminate between men with and men without heart disease. The data is explored by plotting a scattergram of the two variables and estimating their bivariate density function. A scatterplot of the data is constructed in which the members of the no heart disease group are identified by a minus sign and the members of the heart disease group by a plus sign. A first, very crude, estimate of the bivariate density of the two variables can be obtained using the hist2d function in S-PLUS. A far ‘smoother’ estimate of the bivariate density can be obtained by using the kernel estimator approach.
